1892 .
] On the Velocity of Crookes Cathode Stream .
331 to appear , however , will be the longest low-temperature lines of the various chemical elements .
Stage 2.\#151 ; The hydrogen lines will continue to thin out , and the spectra will show many more of the high-temperature lines of different elements .
These will differ from the lines seen in stars of increasing temperature owing to the different percentage composition of the absorbing layers , so far as the known lines are concerned .
Stage 3.\#151 ; With the further thinning out of the hydrogen lines and reduction of temperature of the atmosphere , the absorption flutingsof the compounds of carbon should come in .
that its spectrum shows many of the longest lines of iron .
The conditions at this stage of cooling are satisfied by such stars as / 3 Arietis and x Persei .
In the spectrum of these stars nearly all the solar lines are found , in addition to fairly broad lines of hydrogen .
There is undoubted evidence of the presence of carbon absorption in the solar spectrum and the spectrum of Arcturus , the only star which has yet been investigated with special reference to this point .
The photographs , then , give us the same results as the one formerly obtained from the eye observations .
Comparison is then made betweeu the groups in the classification first suggested by the eye observations , and the various sub-divisions in which the photographs have been arranged .
II .
" On the Velocity of Crookes ' Cathode Stream .
" By Lord Kelvin , P.B.S. Pteceived December 3 , 1892 .
In connection with his splendid discovery of the cathode stream ( stream from the cathode in exhausted glass vessels subjected to electric force ) , Crookes found that when the whole of the stream , or a large part of the whole , is so directed as to fall on 2 or 3 sq .
cm .
of the containing vessel , this part of the glass becomes rapidly heated up to many degrees , as much as 200 ' or 300 ' sometimes , above the temperature of the surroundings .
Let v be the velocity , in centimetres per second , of the cathode stream , and p the quantity of matter of all the molecules in 1 c.c. of it .
Supposing what Crookes ' experiments seem to prove to be not far from the truth , that their impact on the glass is like that of inelastic bodies , and that it spends all their translational energy in heating the glass .
The enero-v thus spent , per square centimetre On the Velocity of Crookes ' Cathode Stream .
[ Dec. 8 of surface struck , per second of time , is 3 ; of which the equivalent in gramme-water-centigrade thermal units is approximately I / ir3/ 42,000,000 .
The initial rate at which this will warm the glass in degrees centigrade per second , is pv io\#174 ; x42T ... ... ... ... ... ... ... ... ... '*)\#187 ; where a denotes the specific heat of the glass , and a the thickness of it at the place where the stream strikes it .
The limiting temperature to which this will raise the glass is 1 -g 3 E X 42,000,000 ... ... ... ... ... ... ... .
( 2 ) where E denotes the sum of the emissivities of the two surfaces of the glass in the actual circumstances .
It is probable that p differs considerably from the average density of the residual air in the enclosure .
Let us take , however , for a conceivably possible example , p \#151 ; 10-8 , which is what the mean density of the enclosed air would be if the vessel were exhausted to 8 x 10~8 of the ordinary atmospheric density .
To complete the example , take v = 100,000 cm .
per sec. ( being about twice the average velocity of the molecules of ordinary air at ordinary temperature ) ; and take era = J- cm .
, as it might be for an ordinary glass vacuum bulb ; and take E \#151 ; which may not be very far from the truth .
With these assumptions , we find , by ( 1 ) and ( 2 ) approximately , 1 ' per second for the initial rise , and 375 ' for the final temperature , which are not very unlike the results found in some of Crookes ' experiments .
The pressure of the cathode stream of the velocity and density which we have assumed by way of example is pv2 , or 100 dynes per square centimetre , or about 100 milligrams heaviness per square centimetre , which is ample for Crookes ' wonderful mechanical results .
The very moderate velocity of 1 kilom .
per second which we have assumed is much too small to show itself by the optical colour test .
.The fact that this test has been applied , and that no indication of velocity of the luminous molecules has been found , has , therefore , no validity as an objection against Crookes ' doctrine of the cathode stream .

